Skip to:

  1. Skip to Jira Navigation
  2. Skip to Side Navigation
  3. Skip to Main Content

[spec] When fspec fails, meaning of ":val" is different than in normal spec failure

Description

Context:

When data fails to conform to a spec, the ":val" of a problem points to the non-conforming data. This has the nice property that the ":val" (for each problem) will exist somewhere within the ":value" (for the entire "explain-data" structure). For example:

(require '[clojure.spec.alpha :as s]) (s/explain-data (s/coll-of int?) [1 2 :a]) ;; #:clojure.spec.alpha{:problems ({:path [], :pred int?, :val :a, :via [], :in [2]}), :spec #object[clojure.spec.alpha$every_impl$reify__934 0x1b235c3e "clojure.spec.alpha$every_impl$reify__934@1b235c3e"], :value [1 2 :a]}

This interpretation of ":val" doesn't seem to apply when a function fails to conform to an fspec spec.

Repro:

(require '[clojure.spec.alpha :as s]) (s/explain-data (s/coll-of (s/fspec :args (s/cat :x int?))) [(fn [%] (/ 1 %))]) ;; #:clojure.spec.alpha{:problems ({:path [], :pred (apply fn), :val (0), :reason "Divide by zero", :via [], :in [0]}), :spec #object[clojure.spec.alpha$every_impl$reify__934 0x420c3355 "clojure.spec.alpha$every_impl$reify__934@420c3355"], :value [#function[expound.alpha/eval28876/fn--28880]]}

Expected: In order to be consistent with the way "val" normally works, "val" should be the anonymous function. While the function arguments are very useful, perhaps they could be associated with a different key?
Actual: "val" contains the arguments to the function, so it's no longer true that the "val" exists within the "value".

Environment

org.clojure/spec.alpha "0.1.134"

Activity

Show:

Ben Brinckerhoff October 30, 2017 at 3:06 PM

Sorry about that - I'll be sure to avoid setting that field in the future.

A little more context: there are cases where the "in" for a problem is tricky to interpret (https://dev.clojure.org/jira/browse/CLJ-2192). In Expound, I have use a heuristic to figure out how the "in" path should work, but that heuristic depends on the "val" existing within the "value". Due to this issue, the heuristic doesn't work, which means I can't reliably report the spec error. See https://github.com/bhb/expound/issues/41

Alex Miller October 30, 2017 at 1:25 AM

Please don't set the fix version - core team will do this when it's targeted to a release. This won't be addressed for 1.9.

Ben Brinckerhoff October 29, 2017 at 9:04 PM

This also makes the error messages misleading:

(s/explain (s/coll-of (s/fspec :args (s/cat :x int?))) [(fn [%] (/ 1 %))]) ;; In: [0] val: (0) fails predicate: (apply fn), Divide by zero

because it seems to indicate that the args are failing to conform, whereas the issue is that the function itself is not conforming.

Details

Assignee

Reporter

Labels

Approval

Triaged

Priority

Affects versions

Created October 29, 2017 at 8:53 PM
Updated June 22, 2018 at 3:32 PM

Flag notifications